Make tSNE for bacteria and virus SGBs, color by metadata

load packages

library(ggplot2)
library(data.table)
library(stringr)
library(dplyr)
library(tibble)
library(tidyr)
library(rprojroot)
library(Rtsne)
library(cowplot)

set paths and filenames

### files
long_table=sprintf("%s/data/mp142_TGVG1.1_MPA4_combined_abundance_table_longform1.tsv", find_rstudio_root_file())
metadata_table=sprintf("%s/data/some_teddy_MP142_metadata2.all_samples1.delivery.csv", find_rstudio_root_file())

load long data table, filter by SGBs detected in over 100 samples

long_dt <- fread(sprintf("%s", long_table), sep = "\t", header = T) %>%
  select(sampleID, rel_abundance, lineage) %>%
  mutate(kingdom = case_when(grepl("k__Bac", lineage) ~ "Bacteria", 
                             grepl("k__Vir", lineage) ~ "Virus",
                             grepl("k__Ar", lineage) ~ "Archea",
                             grepl("k__Euk", lineage) ~ "Eukaryota",
                             TRUE ~ "other")) %>%
  group_by(lineage) %>%
  filter(n() >= 100) %>%
  ungroup()

wide_bac_sp_dt <- long_dt %>%
  filter(kingdom == "Bacteria") %>%
  select(-kingdom) %>%
  pivot_wider(names_from = lineage, values_from = rel_abundance, values_fill = 0)

sampleID_l <- wide_bac_sp_dt$sampleID


wide_bac_sp_dt <- wide_bac_sp_dt %>% select(-sampleID)


wide_vir_sp_dt <- long_dt %>%
  filter(kingdom == "Virus") %>%
  select(-kingdom) %>%
  pivot_wider(names_from = lineage, values_from = rel_abundance, values_fill = 0)

sampleID_x <- wide_vir_sp_dt$sampleID


wide_vir_sp_dt <- wide_vir_sp_dt %>% select(-sampleID)

load metadata table

meta_dt <- fread(sprintf("%s", metadata_table), sep = ",", header = T) %>%
  select(-V1) %>%
  mutate(sample = as.character(sample))

run bacteria PCA

bac_prcomp1 <- prcomp(wide_bac_sp_dt)
names(bac_prcomp1)

[1] "sdev"     "rotation" "center"   "scale"    "x"       

#plot(bac_prcomp1$x[,1:2], pch = ".")

emb_bac <- Rtsne::Rtsne(bac_prcomp1$x[,1:20], perplexity = 20)

embb_bac <- emb_bac$Y

rownames(embb_bac) <- sampleID_l

embb_bac_dt <- as.data.frame(embb_bac) %>%
  setDT(., keep.rownames = "sampleID")

embb_bac_meta_dt <- merge(embb_bac_dt, meta_dt, by.x = "sampleID", by.y = "sample") %>%
  mutate(Country = case_when(
         country == 1 ~ "USA",
         country == 2 ~ "FIN",
         country == 3 ~ "GER",
         country == 4 ~ "SWE",
         TRUE ~ "other"
       ))

bac_dayp <- embb_bac_meta_dt %>%
  ggplot(aes(x=V1, y=V2, color=age_days, shape = Country)) +
  geom_point(size = 1.3, alpha = 0.5) +
  scale_color_gradient2(low = "maroon", mid = "grey", high = "#3F3F7B", 
                        midpoint = 500, limits = c(0,1400), na.value = "black", name = "Day of Life") +
  scale_shape_manual(values=c(15, 16, 17, 18)) +
  theme_bw() +
  labs(x = "tSNE-1", y = "tSNE-2", title = "tSNE of Bacteria in TEDDY samples")
bac_dayp

bac_T1Dp <- embb_bac_meta_dt %>%
  ggplot(aes(x=V1, y=V2, color=T1D, shape = Country)) +
  geom_point(size = 1.3, alpha = 0.5) +
  scale_color_manual(values=c("#8CBEB1", "red")) +
  scale_shape_manual(values=c(15, 16, 17, 18)) +
  theme_bw() +
  labs(x = "tSNE-1", y = "tSNE-2", title = "tSNE of Bacteria in TEDDY samples")
bac_T1Dp

run virus PCA

vir_prcomp1 <- prcomp(wide_vir_sp_dt)

#plot(vir_prcomp1$x[,1:2], pch = ".")

emb_vir <- Rtsne::Rtsne(vir_prcomp1$x[,1:20], perplexity = 20)

emb_vir <- emb_vir$Y

rownames(emb_vir) <- sampleID_x

embb_vir_dt <- as.data.frame(emb_vir) %>%
  setDT(., keep.rownames = "sampleID")

embb_vir_meta_dt <- merge(embb_vir_dt, meta_dt, by.x = "sampleID", by.y = "sample") %>%
  mutate(Country = case_when(
         country == 1 ~ "USA",
         country == 2 ~ "FIN",
         country == 3 ~ "GER",
         country == 4 ~ "SWE",
         TRUE ~ "other"
       ))

vir_dayp <- embb_vir_meta_dt %>%
  ggplot(aes(x=V1, y=V2, color=age_days, shape = Country)) +
  geom_point(size = 1.3, alpha = 0.5) +
  scale_color_gradient2(low = "maroon", mid = "grey", high = "#3F3F7B", 
                        midpoint = 500, limits = c(0,1400), na.value = "black", name = "Day of Life") +
  scale_shape_manual(values=c(15, 16, 17, 18)) +
  theme_bw() +
  labs(x = "tSNE-1", y = "tSNE-2", title = "tSNE of Viruses in TEDDY samples")
vir_dayp

vir_T1Dp <- embb_vir_meta_dt %>%
  ggplot(aes(x=V1, y=V2, color=T1D, shape = Country)) +
  geom_point(size = 1.3, alpha = 0.5) +
  scale_color_manual(values=c("#8CBEB1", "red")) +
  scale_shape_manual(values=c(15, 16, 17, 18)) +
  theme_bw() +
  labs(x = "tSNE-1", y = "tSNE-2", title = "tSNE of Viruses in TEDDY samples")
vir_T1Dp

NA
NA

combine and save

tSNE_combp <- plot_grid(bac_dayp, bac_T1Dp, vir_dayp, vir_T1Dp, align = "h", nrow = 2)

ggsave(tSNE_combp, file = sprintf("%s/charts/tSNE_bacteria_vs_virus_all_samples.pdf", find_rstudio_root_file()), width = 8, height = 8)